About the job
About Fundamental
Fundamental is at the forefront of artificial intelligence, revolutionizing enterprise decision-making. Founded by experts from DeepMind, we have introduced NEXUS – the most advanced Large Tabular Model (LTM) designed specifically for the structured data that underpins critical business decisions. Supported by premier investors and relied upon by Fortune 100 companies, we unlock immense value for businesses, empowering them with predictive capabilities.
In this role, you will tackle unique technical challenges in developing foundational models and create technology that reshapes decision-making processes for some of the largest companies globally. This is a rare opportunity to contribute to a groundbreaking organization from its inception. Join us in defining the future of enterprise AI.
About the Role
Participate in the development and enhancement of a large neural network-based tabular model using Python.
Analyze training and inference pipelines to detect performance bottlenecks.
Refactor essential components in C++ (utilizing PyBind11 or custom extensions) where Python's limitations arise.
Enhance memory efficiency, latency, and throughput throughout model pipelines.
Ensure model correctness, numerical stability, and reproducibility as it evolves.
Collaborate with machine learning researchers to integrate new functionalities into production.
Maintain clear abstractions, robust tests, and thorough documentation.
Contribute to architectural decisions for our machine learning systems managing tabular data.
Qualifications
Solid grounding in software engineering principles with advanced skills in Python and C++.
Practical experience in bridging Python and C++ (PyBind11, Cython, or custom extensions).
Experience in developing and maintaining machine learning models in a production environment.
Deep understanding of neural networks.
Proven ability to optimize performance-critical code.
Strong skills in profiling and debugging (CPU, memory, latency).
Preferred Qualifications
Familiarity with tabular machine learning techniques (transformers, tree/NN hybrids, learned embeddings).
Understanding of PyTorch internals and model optimization techniques.